This is what an industry of coping mechanisms looks like. One attempt to leave Terraform’s model outright was Winglang which compiled a new language down to Terraform and shut down in 2025. When Terraform’s own license changed, the community forked the whole thing rather than let it go. The tool is too entrenched to abandon, so the market lines up to patch it.

The last column is the point. State’s real job is to bind the name you wrote, MyResource, to the opaque id the provider hands back, so the next run knows the two are the same thing. Terraform drives a raw API and catches that id and makes you hold on to it. chant drives no raw API. It compiles to the platform’s own spec, which keeps the binding itself, so there is no pair at all.

The theme is state

The file Terraform keeps has to be hosted and locked so two applies do not corrupt it and then reconciled with a reality that drifts. Every one of those is a gap, and at every gap someone built a product. Remote backends, locking services, state surgery tools, drift detection, a real database to hold it all. It is not the binding of state that seeded this market, but rather the storage requirement.

A fix that needs the problem

A tool that sells you a better store needs it to stay two things at once: bad enough that you will pay to fix it, and permanent enough that you cannot just drop it.

State is corruption, locking, drift, the whole horror show, which is the fear that sells the fix. When someone proposes dropping the store instead of improving it, the same pain gets waved off as solvable.

The severity moves with the argument because the business underneath needs it to be a problem and a fixture in the same breath.

chant keeps no store

chant reads the binding where it already lives. It does not create a resource and catch an id it then has to remember. chant compiles to the platform’s own spec and hands it over. There is no pair to record so there is nothing to host or lock behind it.

This is the same move as honoring the lower layer. Among the tools here it makes chant the only spec-native one. The fair objection is CDK which also compiles to a cloud’s own spec rather than wrapping it. But CDK requires execution, and it speaks a single cloud.

That does not make chant a replacement for the whole industry. It removes one need where most infrastructure lives. What goes away is the state store and everything built to make it bearable.

The measure of a problem

You can size a problem by how many companies exist to soften it. Storing state yourself has produced a lot of them. Fork the license or wrap the CLI or host the store or graph it or leave the model behind. All of it follows from driving a raw API and catching the id yourself.

The cheapest state store is the one you never have to keep.

Read more

This post mentioned Terragrunt · TerraConstructs · Atlantis · Spacelift · env0 · Terraform Cloud · Stategraph · compliance.tf · Winglang